roberta-polish-kgr10 / config.json
Łukasz Augustyniak
first version of clarin poslish roberta based on kgr10 dataset
2c5669c
raw
history blame
567 Bytes
{
"architectures": [
"RobertaForMaskedLM"
],
"attention_probs_dropout_prob": 0.1,
"bos_token_id": 0,
"eos_token_id": 2,
"gradient_checkpointing": false,
"hidden_act": "gelu",
"hidden_dropout_prob": 0.1,
"hidden_size": 768,
"initializer_range": 0.02,
"intermediate_size": 3072,
"layer_norm_eps": 1e-12,
"max_position_embeddings": 514,
"model_type": "roberta",
"num_attention_heads": 12,
"num_hidden_layers": 12,
"pad_token_id": 1,
"position_embedding_type": "relative_key_query",
"type_vocab_size": 1,
"vocab_size": 52000
}